From: Robert <bob.news@...>
Date: 2003-02-27T16:18:24+09:00
Subject: Re: Ruby Compile-time optimization

^Hi,

did you do this with a loop body that does really do something?  I mean,
your example is a bit artificial and optimizing this scenario would have not
much impact for real code, because nobody would repeat a constant operation
in a loop exept for performance measurements.

You could simply sum up a sequence of random numbers or something similar.

Kind regards

    robert


"Daniel Carrera" <dcarrera@math.umd.edu> schrieb im Newsbeitrag
news:20030227002438.GA3877@math.umd.edu...
> Hello,
>
> I was wondering about what exactly makes Perl faster than Ruby, and I made
> a small test to examine this a bit.  I sort of assumed that it was because
> Ruby has more powerful data structures, but it seems that compile-time
> optimization might be a significant factor.
>
> The first step of my test was to look at the loop speed of Perl and Ruby.
> I ran a loop that did nothing at all about 500,000 times.  These are the
> average of 3 tries:
>
> ruby      1.023s
> perl      1.140s
>
> So Ruby's loop was faster than Perl's.  I then put 'val = 2 + 2' inside
> each loop (average of 3 tries):
>
>         Total    Minus-the-loop
> ruby    2.660s   1.637s     100.0%
> perl    1.770s   0.630s      38.5%
>
>
> In the second column I subtract the loop time from the total to get the
> time spent in the "val = 2 + 2" step.  Here Perl is better than twice as
> fast as Ruby.
>
> Now, as many of you know, Perl is not exactly an interpreted language.
> Perl first "compiles" the program into some internal bytecode, which is
> optimized before being run.  For instance, if you have this:
>
> use constant  $MYCONST => 3
> $val = 2 + $MYCONST
>
> Perl turns it into:
>
> $val = 5
>
> I ran the ruby code again, but this time I replaced 'val = 2 + 2' by 'val
> = 4' to simulate the effect of optimization.  Again, the average of 3
> tries:
>
>                  Total     Minus-the-loop
> ruby             2.660     1.637     100.0%
> ruby -optimized  1.793     0.770      47.0%
> perl             1.770     0.630      38.5%
>
> Now ther Perl's operation is only about 18% faster, which in this case
> particular case is mitigated by Ruby's loop being 11% faster.  Overall,
> the Perl code is only 1% faster (which is less than experimental error).
>
> I realize that the fact that Ruby is a more dynamic language makes
> optimization more difficult, but I really think that we should look into
> it.
>
> Just a thought.
> --
> Daniel Carrera
> Graduate Teaching Assistant.  Math Dept.
> University of Maryland.  (301) 405-5137
>